Male Sprague Dawley rats (total n = 133; 4 independent cohorts) (Charles River; Quebec, Canada) arrived at postnatal day (PND) 28. Rats were group housed (2–3 rats per cage) at standard conditions (temperature: +24 ± 2 °C; humidity: 55 ± 10%; light schedule: 12:12 h, light on at 7AM) with food and water available ad libitum unless stated otherwise. Rats were assigned to experimental groups randomly at the beginning of adolescent treatment. All procedures were approved by the Institutional Animal Care Committee and complied with the Canadian Council on Animal Care guidelines.

Adolescent THC treatment was conducted between PND35 and PND45 (Fig. 1a). Rats were treated with increasing doses of THC as described previously [5]: 2.5 mg/kg (PND35-37), 5 mg/kg (PND38-41) and 10 mg/kg (PND42-45) or with vehicle. For NAC intervention experiments, rats were co-treated with N-acetylcysteine (NAC, Sigma Aldrich; Cat#A7250) administered ad libitum in drinking water (900 mg/l; PND35-60) and fresh solution was prepared every 2 days. Cremophor EL (Sigma Aldrich) was added to THC-EtOH solution (1 g/100 ml EtOH; Cayman Chemicals), vortexed and EtOH was evaporated using nitrogen stream. Subsequently, the THC solution was diluted in physiological saline to a final concentration of 2.6 mg/ml and 5% Cremophor. Vehicle contained 5% Cremophor in saline.Fig. 1

a Schematic illustration for experimental treatment timelines. b Acute effects of adolescent THC exposure on body weight are not impacted by NAC co-treatment (n: VEH = 38; THC = 40; VEH-NAC = 24; THC-NAC = 24). Repeated measures two-way ANOVA: PND day: F_(4,488) = 2301.214, p < 0.001; PND day*IP treatment: F_(4,488) = 29.413, p < 0.001; PND day*oral treatment: F_(4,488) = 0.07, p = 0.991; PND day*IP*oral treatment: F_(4,488) = 1.169, p = 0.324; pairwise comparisons VEH vs. THC at PND37-45: p < 0.001; VEH-NAC vs. THC-NAC at PND37: p = 0.014, PND39: p = 0.003, PND41: p < 0.001, PND43: p = 0.003, PND47: p = 0.001). c Example motor activity traces for each treatment group with the center area highlighted with pale orange background. Rats in all treatment groups (n: VEH = 18, THC = 18, VEH-NAC = 24; THC-NAC = 24) entered the center of the arena with similar frequency d and duration e. Although there were no differences between groups in total activity levels f, THC treatment significantly increased stereotyped rearing frequency g (two-way ANOVA: IP treatment: F_(1,80) = 4.19, p = 0.044; oral treatment: F_(1,80) = 0.509, p = 0.47; IP*oral treatment: F_(1,80) = 3.77, p = 0.056; pairwise comparisons: VEH vs. THC: p = 0.01). Significance indicators on the right point significance level for main effects observed between VEH vs. THC (*) and VEH-NAC vs. THC-NAC (#). Data are presented as mean± S.E.M. with individual data points superimposed on bar graphs, for this and all subsequent figures. Asterisks = *p < 0.05, **p < 0.01, ***p < 0.001 for this and all subsequent figures.

All behavioural experiments were conducted on adult animals >PND75, video recorded and analyzed offline using Behaview software (www.pmbogusz.net) and Any-maze (San Diego Instruments, RRID:SCR_014289) software. Experiments were analyzed in a blinded manner and all behavioural tests were independently replicated at least once. Sample sizes per experimental group were chosen based on previously reported similar studies [4, 5].

Following behavioural experiments, animals were implanted with extracellular recordings electrodes under deep ketamine-xylazine anaesthesia (i.p.; ketamine: 80 mg/kg, Vetoquinol; xylazine: 6 mg/kg, Bayer) using a sterile stereotaxic procedure. Electrodes were made of polyimide coated stainless-steel wire (120 μm; WPI) secured in the electrode pedestal (MS636, PlasticsOne) and targeted mPFC: +3.2 mm anterior and −0.8 mm lateral from Bregma, and 3.2 mm ventral to cortical surface. A stainless-steel bone screw was implanted above the cerebellum and served as the grounding and reference electrode. The implant was secured in place with 3 additional bone screws and dental cement. Rats were recovering for 7 days before recordings. Local field potential signals (LFP) were acquired at a 1 kHz rate using RZ6 processor (TDT) and band-pass filtered between 0.5–300 Hz. Data were analyzed using custom made scripts in Matlab.

Patch-clamp experimental procedures are detailed in the Supplementary Materials and Methods. Briefly, adult rats (PND75-PND130) with histories of adolescent THC vs. VEH treatments were anesthetised with isoflurane and decapitated (n = 5–8 animals/group). Brains were quickly extracted and sliced (300 μm) for whole cell recordings in current-clamp and voltage-clamp mode.

Rats were given an overdose of pentobarbital (Euthanyl; 270 mg/ml), and brains were removed and flash frozen (–80 °C). For quantification of multiple neurotransmitters and metabolites two different matrices were used, zinc oxide nanoparticles (ZnO-NP, Sigma-Aldrich, St. Louis, MO) and 4-(Anthracen-9-yl)-2-fluoro-1-methylpyridin-1-ium iodide (FMP-10, TAG-ON, Uppsala, Sweden; Cat#T1001) and are detailed in the Supplementary Materials and Methods.

All data are presented as mean ± standard error of the mean (S.E.M.) and were analyzed using SPSS software (IBM; RRID:SCR_016479) unless stated otherwise. Normally vs. non-normally distributed data (Kolmogorov-Smirnoff test; p < 0.05) were tested using appropriate parametric vs. non-parametric analysis followed with post-hoc tests, where appropriate. The data sets were tested with either two-way ANOVA or repeated measures two-way ANOVA with adolescent IP treatment and oral treatment set as fixed factors and repeated measures set to the following: postnatal day (for detecting body weight changes during the treatment), pulse intensity (for assessing startle reactivity), pre-pulse intensity (for assessing sensory-motor gating in PPI test) and injected current amplitude (for detecting excitability changes in patch-clamp experiment). ANOVAs were followed with Sidak-corrected, post-hoc pairwise comparisons. Two-tailed paired t-tests were used to detect within-group differences between exploration times in social behaviour, object recognition tests and habituation of the startle response.

Adolescent THC exposure induces anxiogenic phenotypes and increases stereotypy behaviours in the open field task at adulthood [4, 5, 10]. Stereotypy is commonly observed in schizophrenia, characterized by repetitive, perseverative motor behaviours associated with negative symptom clusters [45]. We found that no treatment conditions affected locomotion or thigmotaxis measured in the OF test (Fig. 1c–f), however THC rats displayed increased stereotypy with significantly increased rearing counts vs. controls. NAC co-treatment prevented this THC effect (Fig. 1g), though rearing counts in both NAC treatment groups were elevated compared to VEH controls. Importantly, a recent study has shown that NAC treatment was not able to prevent THC-induced impairments in anxiety [46], suggesting that NAC might be less efficient in mitigating affective impairments that rely on amygdala function.

Adolescent THC exposure impairs social behaviours at adulthood [4, 5, 10] and clinically, such deficits are well-established ‘negative’ symptom endophenotypes of schizophrenia. We used a social motivation/memory test to determine if NAC may prevent these deficits (Fig. 2a, e). Baseline sociability was similar across all cohorts with rats preferring to explore novel rat vs. empty enclosures (Fig. 2b). Social preference scores were uniform across treatments (Fig. 2c) and the percentage of rats showing preference toward the novel rat (score >0.55) was similar (Fig. 2d). However, in phase 2 of the test THC-exposed rats exhibited significantly impaired social memory, failing to distinguish between the novel vs. familiar rat (Fig. 2f) and displayed lower recognition vs. VEH and THC-NAC treated rats (Fig. 2g). Consistently, the percentage of animals preferring the novel rat was lower in the THC group (Fig. 2h). Thus, NAC prevented the deleterious impact of adolescent THC-exposure on social recognition memory.

We first examined the effects of NAC on THC-induced impairments in the object recognition memory (NOR; Fig. 2i) task. During the acquisition phase both objects were explored similarly (paired t-tests for left vs. right object: p’s>0.05, data not shown) indicating no environmental bias. Following, one object was substituted and significant preference toward novelty was detected in all groups, except in THC exposed rats (Fig. 2j). This was reflected by significantly reduced object recognition scores in the THC group (Fig. 2k) and higher percentage of rats displaying no novelty preference (recognition score <0.55; Fig. 2m). Importantly, NAC supplementation reversed this deficit in THC exposed cohorts (Fig. 2k, m). Next, rats underwent the temporal order test (TMOR; Fig. 2n), which exploits rats’ preference to explore objects encountered earlier vs. more recently. During both acquisition phases, there were no side preferences observed (paired t-tests p’s > 0.05, data not shown). At the retention test, only THC exposed rats showed impaired recent object recognition (Fig. 2o), lower recognition score (Fig. 2p) and a higher percentage of rats with a recognition score <0.55 (Fig. 2r). Overall exploration times were homologous (Fig. 2q), confirming that the decreased temporal recall resulted from cognitive impairment and not from a reduced exploration. Finally, to determine if THC-induced memory deficits might be related to perceptual deficits, we employed the simultaneous oddity discrimination task (Supplemental Fig. 1a) which relies on the natural preference of rats to explore odd vs. similar objects. Analyses revealed no significant group differences across treatments in total exploration times or in oddity discrimination scores (Supplemental Fig. 1b, c), demonstrating that adolescent treatment with THC or NAC is not impacting perceptual capacity in adulthood. Summarizing, NAC prevented THC-induced object memory deficits without interfering with object recognition memory when applied alone.

Given the importance of the PFC in attentional and cognitive flexibility [47], we examined whether NAC would prevent THC-induced deficits in these domains. We used a set-shifting task [47] that assess attentional flexibility, modelled after the Wisconsin Card Sorting Test used in schizophrenia patients, who display impaired reversal learning in this task [48]. First, rats were trained to press a lever below an illuminated cue light for a food reward (Fig. 3a). Successful training occurred in all treatment groups with NAC treatment alone improving acquisition speed (Fig. 3b). Testing visual discrimination memory retrieval revealed no THC effects, while NAC co-treatment enhanced it (Fig. 3c), indicating memory persistence in all groups and intact consolidation processes. Next, rats were required to ignore the cue light and always press the same lever (set-shifting; Fig. 3d). There were no differences between groups and all groups switched to the new rule (Fig. 3e), retrieving the memory the following day (Fig. 3f). Changing the rule again (reversal learning, Fig. 3g), significantly increased the number of trials (Fig. 3h) and errors (Fig. 3i) selectively in THC-treated rats. As evident from the average cumulative correct responses across the session (Fig. 3j), learning in THC animals was delayed when compared to VEH rats. Moreover, the percentage of animals that did not reverse was elevated only in the THC cohort (Fig. 3k). Thus, adolescent antioxidant treatment with NAC is sufficient to prevent the deleterious side-effects of adolescent THC exposure on cognitive flexibility.Fig. 3

a, d, g Schematics showing experimental design of the extra-dimensional set-shifting task. Rats were trained to press the lever above which a cue light was illuminated a. b NAC-treated rats acquired the association between pressing the illuminated lever and obtaining food reward faster than those drinking water (n: VEH = 14, THC = 16, VEH-NAC = 8; THC-NAC = 7). Two-way ANOVA: IP treatment: F_(1,41) = 1.057, p = 0.31; oral treatment: F_(1,41) = 9.682, p = 0.003; IP*oral treatment: F_(1,41) = 0.03, p = 0.863; pairwise comparisons: VEH vs. VEH-NAC: p = 0.04; THC vs. THC-NAC: p = 0.028. c Next day, the recall of this learned association was improved with NAC treatment vs. THC exposed groups: Two-way ANOVA: IP treatment: F_(1,41) = 0.551, p = 0.462; oral treatment: F_(1,41) = 8.571, p = 0.006; IP*oral treatment: F_(1,41) = 0.244; p = 0.624; pairwise comparisons: VEH vs. VEH-NAC: p = 0.086; THC vs. THC-NAC: p = 0.023. d Next, rats were required to press the lever on one side of the cage irrespective of the cue light. e The number of trials required to shift the strategy did not differ between groups. Two-way ANOVA: IP treatment: F_(1,41) = 0.21, p = 0.649; oral treatment: F_(1,41) = 0.001, p = 0.97; IP*oral treatment: F_(1,41) = 0.773, p = 0.384. f Set-shifting memory recall the following day was not affected by any treatment condition: Two-way ANOVA: IP treatment: F_(1,41) = 1.774, p = 0.19; oral treatment: F_(1,41) = 1.067, p = 0.308; IP*oral treatment: F_(1,41) = 0.777, p = 0.383. g However, changing the rule again to the opposite lever resulted in h increased numbers of trials required for the cognitive shift and i more errors committed by THC-exposed rats: Two-way ANOVA’s: #of trials: IP treatment: F_(1,41) = 2.037, p = 0.161; oral treatment: F_(1,41) = 1.426, p = 0.239; IP*oral treatment: F_(1,41) = 5.294, p = 0.027; pairwise comparisons: VEH vs. THC p = 0.002; THC vs. THC-NAC p = 0.02; # of errors: IP treatment: F_(1,41) = 2.284, p = 0.138; oral treatment: F_(1,41) = 5.177; p = 0.028; IP*oral treatment: F_(1,41) = 7.002, p = 0.011; pairwise comparisons: VEH vs. THC p < 0.001; THC vs. THC-NAC p = 0.001. j Average performance of rats during the reversal learning. Note the rightward shift in THC-exposed rats, reflecting more erroneous responses. This pathological error shift is prevented by NAC treatment. k Schematic summarizing the % of rats that successfully acquired the new cognitive strategy tactic. Note the profound deficit in THC treated cohorts and prevention in the NAC-treated cohorts.

Adolescent THC exposure leads to sensorimotor gating impairments manifesting as disrupted prepulse inhibition (PPI) of startle, a cardinal endophenotype of schizophrenia [49]. Thus, we examined the potential effects of NAC to prevent these effects. Assessment of the acoustic startle response to increasing tone intensities showed exaggerated acoustic reactivity in THC treated rats (Fig. 4a–c). Next, we assessed sensorimotor gating by measuring PPI (Fig. 4d). NAC and THC-NAC rats showed PPI similar to VEH controls, whereas THC rats had reduced PPI across multiple prepulse intensities (Fig. 4e). Lastly, we tested short-term startle habituation (Fig. 4f). In contrast to all other groups, the final vs. initial startle responses of THC group did not significantly differ (Fig. 4g) and startle habituation levels were significantly lower compared to VEH and THC-NAC rats (Fig. 4h). Accordingly, the percentage of rats displaying at least a 15% decrease of startle amplitude was lower only in the THC group (Fig. 4i), demonstrating that NAC protects against the deleterious THC effects on adulthood sensorimotor gating.Fig. 4

a Schematic representation of startle apparatus. b THC-exposed rats display increased startle reactivity (data presented as mean ± SEM; n: VEH = 9, THC = 9, VEH-NAC = 11; THC-NAC = 12). Two-way repeated measures ANOVA: Within-subjects effects: startle amplitude: F_{(10, 370)} = 121.994, p < 0.001; startle amplitude*IP treatment: F_{(10, 370)} = 2.514, p = 0.006; Startle amplitude*oral treatment: F_{(10, 370)} = 0.57, p = 0.838; startle amplitude*IP*oral treatment: F_{(10, 370)} = 0.609, p = 0.807; between-subjects effects: IP treatment: F_(1,37) = 0.002; oral treatment: F_(1,37) = 1.258; p = 0.269; IP*oral treatment: F_(1,37) = 1.379, p = 0.248; pairwise comparisons: pulse intensity: VEH vs. THC: 65 dB p = 0.026; 90 dB p = 0.025; 100 dB p = 0.007; 105 dB p = 0.009. c Depiction of startle response magnitudes across individual animals (rows within the squares) to startle pulse presentation of increasing intensity (discrete pixels; 65–115, 5 dB increment; lowest intensity at the leftmost position of the squares). The startle magnitude is color coded with inset below the graph. d Schematic depiction of PPI protocol. e Reduced PPI in THC-exposed rats was reversed with NAC. Two-way RM ANOVA: tests of within-subjects effects: startle inhibition: F_(2,74) = 50.099; p < 0.001; between-subjects effects: IP treatment: F_(1,37) = 1.2, p = 0.28; oral treatment: F_(1,37) = 5.698; p = 0.022; IP*oral treatment: F_(1,37) = 5.768; p = 0.021; pairwise comparisons for means: VEH vs. THC: p = 0.025; THC vs. THC-NAC: p = 0.002, pairwise comparisons for prepulse intensity: VEH vs. THC at 77 dB: p = 0.017, at 80 dB: p = 0.058, at 83 dB: p = 0.19; THC vs. THC-NAC: at 77 dB: p = 0.005; at 80 dB: p = 0.009; at 83 dB: p = 0.007. f Schematic of short-term startle habituation protocol and example depicting gradual decrease of startle responses. g Paired t-test revealed that median startle amplitude from last ten trials did not differ from first three trials in the THC group (VEH: t₍₈₎ = 3.988, p = 0.004; THC: t₍₈₎ = −0.329, p = 0.751; VEH-NAC: t₍₁₀₎ = 4.897, p < 0.001; THC-NAC: t₍₁₁₎ = 4.806, p < 0.001). h The habituation score remained high only in THC group. Two-way ANOVA: IP treatment: F_(1,37) = 5.586, p = 0.023; oral treatment: F_(1,37) = 4.684, p = 0.037; IP*oral treatment: F_(1,37) = 6.869, p = 0.013; pairwise comparisons: VEH vs. THC: p = 0.002; THC vs. THC-NAC: p = 0.002. i Radial plot capturing the inverted percentage of animals with impaired startle habituation in the THC group.

Adolescent THC exposure causes enduring mPFC neuronal abnormalities, including hyperactivity of pyramidal neurons in vivo and increased gamma power of oscillatory potentials [4, 5]. We next conducted ex vivo whole-cell patch-clamp recordings from layer 2/3 pyramidal neurons from the adult mPFC. THC treatment did not affect passive membrane properties relative to VEH (Fig. 5a–c), and action potential properties were similar across groups (Fig. 5d, e), except for enlarged spike afterhyperpolarization amplitude upon NAC treatment (Fig. 5f). There was a trend towards augmented neuronal gain in response to increasing depolarizing current injection in the THC group (Fig. 5g, h) and a significant increase of spiking upon a depolarizing ramp (Fig. 5i–k) along with a trend toward lower rheobase values (Supplemental Fig. 2). Furthermore, the frequency but not amplitude of spontaneous excitatory postsynaptic currents (sEPSCs) in THC rats was significantly higher indicating escalation of excitatory synaptic input. Although NAC prevented this frequency shift (Fig. 5l–p) it increased sEPSCs amplitude. Lastly, we measured presynaptic changes by employing paired-pulse ratio (PPR) and we detected no effects of adolescent treatments on evoked excitatory postsynaptic currents (ePSCSs; Fig. 5q, r).Fig. 5

a Input resistance, b membrane time constant and c resting membrane potential were not affected by adolescent THC treatment (n: VEH = 29, THC = 27, VEH-NAC = 15; THC-NAC = 20). Two-way ANOVAs: a IP treatment: F_(1,87) = 2.031,p = 0.158; oral treatment: F_(1,87) = 0.297, p = 0.587; IP*oral treatment: F_(1,87) = 0.064, p = 0.8. b IP treatment: F_(1,87) = 0.176, p = 0.676; oral treatment: F_(1,87) = 6.174, p = 0.015; IP*oral treatment: F_(1,87) = 1.252, p = 0.266; pairwise comparisons: THC vs. THC-NAC p = 0.01. c IP treatment: F_(1,87) = 1.477, p = 0.227; oral treatment: F_(1,87) = 2.586, p = 0.111; IP*oral treatment: F_(1,87) = 0.789, p = 0.377. d Spike half-width and e spike threshold also did not differ between groups. Two-way ANOVAs: d IP treatment: F_(1,87) = 0.335, p = 0.564; oral treatment: F_(1,87) = 3.288, p = 0.073; IP*oral treatment: F_(1,87) = 0.704, p = 0.404; e IP treatment: F_(1,87) = 0.009, p = 0.926; oral treatment: F_(1,87) = 3.075, p = 0.083; IP*oral treatment: F_(1,87) = 0.138, p = 0.711. f Spike AHP was augmented by NAC treatment. Two-way ANOVA: IP treatment: F_(1,87) = 0.023, p = 0.88; oral treatment: F_(1,87) = 10.52, p = 0.002; IP*oral treatment: F_(1,87) = 0.107, p = 0.745; pairwise comparisons: VEH vs. VEH-NAC p = 0.05; THC vs. THC-NAC p = 0.01. g Response of representative layer 2/3 mPFC pyramidal neurons obtained from different treatment groups to current step of +230 pA (square pulse illustrated below voltage trace). h Neuronal gain curves indicate a trend toward increase in excitability following THC exposure, although no interactions with treatments were detected. Two-way repeated measures ANOVA: depolarization step: F_(5,435) = 244.787, p < 0.001; depolarization step*IP treatment: F_(5,435) = 1.817, p = 0.108; depolarization step*oral treatment: F_(5,435) = 0.556, p = 0.734; depolarization step*IP*oral treatment: F_(5,435) = 1.526, p = 0.180. i Example traces illustrating action potential firing in response to a current ramp (ramping current injection from 0–400 pA in 1.5 s; 0.266 pA/ms). j Average no. of spikes/ramp was significantly increased following THC exposure. Two-way ANOVA: IP treatment: F_(1,85) = 5.197, p = 0.025; oral treatment: F_(1,85) = 2.251, p = 0.137; IP*oral treatment: F_(1,85) = 3.874, p = 0.052; pairwise comparisons: VEH vs. THC p = 0.001; THC vs. THC-NAC p = 0.013. k Percentage of cells with more/less than 10 spikes in a ramp across treatment groups. l sEPSCs frequency was increased in cells from THC-exposed rats (n = 15–21/group). Two-way ANOVA: IP treatment: F_(1,67) = 3.039, p = 0.086; oral treatment: F_(1,67) = 10.265, p = 0.002; IP*oral treatment: F_(1,67) = 6.475, p = 0.013; pairwise comparisons: VEH vs. THC p = 0.002; THC vs. THC-NAC p < 0.001. m Percentage of cells with sEPSC freq< or >2 Hz across different groups (n: VEH = 17, THC = 21, VEH-NAC = 15; THC-NAC = 16). n sEPSCs amplitude was increased in cells from NAC exposed rats, but not affected by THC exposure. Two-way ANOVA: IP treatment: F_(1,67) = 0.497, p = 0.483; oral treatment: F_(1,67) = 15.118, p < 0.001; IP*oral treatment: F_(1,67) = 0.269, p = 0.606; pairwise comparisons: VEH vs. VEH-NAC: p = 0.022; THC vs. THC-NAC: p = 0.002. o Percentage of cells with average sEPSCs amplitude smaller/larger than 10 mV across treatments. p Examples of sEPSCs recorded from adult pyramidal neurons following different adolescent treatments. q Paired-pulse facilitation ratio is not affected by adolescent treatment (n: VEH = 11, THC = 9, VEH-NAC = 15; THC-NAC = 17). Two-way RM ANOVA: stimulation interval: F_(3,144) = 6.805,p < 0.001; stimulation interval*IP treatment: F_(3,144) = 0.832, p = 0.478; stim interval*oral treatment: F_(3,144) = 0.172, p = 0.915; Stim interval*IP*oral treatment: F_(3,144) = 1.311, p = 0.273. r Example recordings representing EPSCs evoked by stimulation pulses delivered with 20, 50, 100 and 200 ms interval (from top to bottom). s Schematic depiction of LFP recording with an example raw signal trace and power spectrum analysis (inset). t and u Average normalized power for specific oscillatory bands was affected by adolescent treatments (n: VEH = 8, THC = 8, VEH-NAC = 6; THC-NAC = 7). Two-way ANOVAs for delta: IP treatment: F_(1.25) = 2.523, p = 0.125; oral treatment: F_(1.25) = 23.012, p < 0.001; IP*oral treatment: F_(1.25) = 1.919, p = 0.178; pairwise comparisons: VEH vs. THC: p = 0.035; VEH vs. VEH-NAC: p = 0.026; THC vs. THC-NAC: p < 0.001; theta: IP treatment: F_(1.25) = 0.109, p = 0.744; oral treatment: F_(1.25) = 15.214, p < 0.001; IP*oral treatment: F_(1.25) = 0.533, p = 0.472; pairwise comparisons: VEH vs. VEH-NAC: p = 0.038; THC vs. THC-NAC: p = 0.003; alpha: IP treatment: F_(1.25) = 1.428, p = 0.243; oral treatment: F_(1.25) = 25.1, p < 0.001; IP*oral treatment: F_(1.25) = 0.076, p = 0.785; pairwise comparisons: VEH vs. VEH-NAC: p = 0.003; THC vs. THC-NAC: p < 0.001; beta: IP treatment: F_(1.25) = 2.449, p = 0.13; oral treatment: F_(1.25) = 14.204, p < 0.001; IP*oral treatment: F_(1.25) = 0.083, p = 0.776; pairwise comparisons: VEH vs. VEH-NAC: p = 0.024; THC vs. THC-NAC: p = 0.007; gamma: IP treatment: F_(1.25) = 4.432, p = 0.045; oral treatment: F_(1.25) = 6.456, p = 0.018; IP*oral treatment: F_(1.25) = 4.784, p = 0.038; pairwise comparisons: VEH vs. THC: p = 0.004; THC vs. THC-NAC: p = 0.002. Patch clamp recordings were taken from layer 2/3 mPFC pyramidal neurons obtained from 28 adult male rats (VEH = 7, THC = 8, VEH-NAC = 5; THC-NAC = 8).

To further explore the impact of excitability changes at the circuit level, we examined local field potentials (LFP) in freely behaving rats (Fig. 5s). Prior work has identified increased mPFC gamma power in anesthetized rats following adolescent THC exposure [5]. Consistently, we observed that adolescent THC exposure significantly increased gamma-power, concomitant with an impaired excitatory-inhibitory balance. Although the excessive THC-induced contribution of gamma band oscillations was reversed by antioxidant co-treatment, NAC also had significant effects on slower oscillatory bands, namely, by increasing delta and reducing theta, alpha and beta powers relative to VEH controls (Fig. 5t, u).

Together, these results demonstrate that adolescent THC exposure has no effect on passive membrane properties of PFC pyramidal cells, however, it significantly increases excitability in response to depolarization and spontaneous excitatory synaptic transmission, which further manifested in increased gamma power in behaving animals, consistent with previously reported hyperactivity of PFC neurons and dysregulation of γ-oscillations [4, 5]. Importantly, NAC treatment normalized both the frequency of EPSCs and γ-band power contributions.

Previous studies reported that adolescent THC exposure leads to persistent changes in various molecular signalling pathways in the mPFC and associated neurotransmitter profiles associated with DA, GABA and Glu [4, 5]. Accordingly, we next used MALDI imaging (Fig. 6a, b) to measure, across treatment groups, the relative abundance of multiple neurotransmitters: glutamate (Glu), glycine (Gly), γ-amino-butyric acid (GABA), dopamine (DA), noradrenaline (NA) and serotonin (5-HT); their metabolites/precursors: 3,4-dihydroxyphenylacetaldehyde (DOPAL), 3,4-dihydroxyphenylacetic acid (DOPAC) and homovanillic acid (HVA); and neuroactive amino acids and molecules: aspartate (Asp), arginine (Arg), alanine (Ala), N-acetylaspartate (NAA), tyramine, taurine, creatinine, creatine, spermine and spermidine (Fig. 6c, d). The average levels (Fig. 6e, f & Supplemental Fig. 3) of Glu and NA remained unaffected by treatments, whereas GABA and 5-HT were significantly decreased in THC and THC-NAC groups (Fig. 6i, j). Amongst other metabolites, significant effects were observed for Arg (increased in THC-NAC treated vs. VEH); taurine (decreased in THC and THC-NAC treated groups), creatinine (decreased in THC and VEH-NAC treated groups vs. VEH) and spermidine (decreased in VEH-NAC and THC-NAC treated groups). Finally, DA was significantly increased in THC rats, but not in NAC and THC-NAC treated rats. Exposure to THC-NAC significantly reduced HVA levels (Fig. 6g, h, k, l).Fig. 6

Optical images of coronal mPFC sections using ZnO-NP a or FMP-10 matrices b, followed by ion distribution plots for selected molecules. Metabolites were simultaneously assessed in a single scan for each matrix, and images are from a single VEH rat. The identity of visualized molecules and corresponding experimental m/z values are indicated above the plots. At each sampling position, 40 shots were used to acquire data in the m/z 50-500 (ZnO-NP) and m/z 299–1000 (FMP-10) range. Ion intensity data were visualized using a color scale 0–100% (inset on the right) except for 5-HT (0–80%); NA (0–75%); spermine (0–60%); DA, DOPAL, DOPAC, Arg, Asp (0–50%), Gln (0–30%) for best visualization. Lateral resolution, 100 µm. Average mass spectra from the mPFC region (delineated in optical images in a and b) facilitated by derivatization with ZnO-NP (blue; c) and FMP10 (crimson; d) with mass peaks for selected molecules displayed on insets. e Box plots of average log2 transformed ratios (ratio = AUC treatment/AUC VEH) for Glu, Arg, GABA, 5-HT, DA, HVA, taurine, creatinine and spermidine and f summary color plot of all assessed molecules showing the decrease (blue) or increase (pink) in the ion abundances across treatment groups relative to VEH (n = 16/group). Molecule identities are listed on the left side of the graph, where red colors indicate significant (p < 0.05) effect of statistical comparisons (2-way ANOVA followed with pairwise comparisons or Kruskal-Wallis test followed with Mann-Whitney tests). Significant post-hoc between-groups comparisons (*vs. VEH, ^ vs. THC, # vs. VEH-NAC) are indicated with: single (p < 0.05), double (p < 0.01) or triple symbol (p < 0.001) respectively. Arg (2-way ANOVA: IP treatment: F_(1,60) = 0.035, p = 0.853; oral treatment: F_(1,60) = 7.474, p = 0.008; IP*oral treatment: F_(1,60) = 1.368, p = 0.247; pairwise comparisons: THC-NAC vs. THC p = 0.008). GABA (Kruskal-Walis: H₍₃₎ = 22.18, p < 0.001; THC vs. VEH: U = 64, p = 0.015, THC-NAC vs. VEH: U = 2, p < 0.001, THC-NAC vs. VEH-NAC: U = 40, p = 0.001). 5-HT (Kruskal-Walis: H₍₃₎ = 13.05, p = 0.005; THC vs. VEH: U = 65, p = 0.017; THC-NAC vs. VEH: U = 42, p = 0.001, THC-NAC vs. VEH-NAC: U = 59, p = 0.008). DA (Kruskal-Walis H₍₃₎ = 8.877, p = 0.031; THC vs. VEH: U = 62, p = 0.012, THC vs. VEH-NAC: U = 68, p = 0.023, THC vs. THC-NAC: U = 66, p = 0.019). HVA (K-W: H₍₃₎ = 11.731, p = 0.008; VEH-NAC vs. VEH: U = 69, p = 0.026, THC-NAC vs. VEH: U = 32, p < 0.001). Taurine (K-W H₍₃₎ = 9.612, p = 0.022; THC vs. VEH: U = 68, p = 0.023; THC-NAC vs. VEH: U = 58; p = 0.007). Creatinine (2-way ANOVA: IP treatment: F_(1,60) = 8.807, p = 0.004; oral treatment: F_(1,60) = 6.634, p = 0.012; IP*oral treatment: F_(1,60) = 2.075, p = 0.155; THC vs. VEH p = 0.003; VEH-NAC vs. VEH p = 0.006). Spermidine (K-W test H₍₃₎ = 14.152, p = 0.003; VEH-NAC vs. VEH: U = 50, p = 0.009; THC-NAC vs. VEH: U = 29, p < 0.001; THC-NAC vs. THC: U = 68, p = 0.041; THC-NAC vs. VEH-NAC: U = 62, p = 0.037).

Oral NAC is associated with gastrointestinal discomfort, nausea, and diarrhea, but these symptoms are generally mild and occur at similar rates in placebo groups, indicating that NAC is well-tolerated and safe for chronic use [108]. In our study, we did not observe an increase in loose stools in NAC-treated animals during daily monitoring, consistent with previous research showing no toxicity after 30 days of high-dose NAC (1200 mg/kg/day) in rats [109]. After oral administration, NAC is primarily metabolized in the liver to cysteine, which enters the bloodstream [110]. Both NAC and cysteine have limited ability to cross the blood-brain barrier (BBB) [111]. Despite this, NAC administration increases GSH levels in rats [112, 113] and humans [114]. Transport across the BBB may be facilitated by sodium-dependent transport system [76] or converting cysteine to cystine, which is subsequently exchanged with intracellular glutamate via the cystine-glutamate antiporter, and reduced back to cysteine for GSH production [115, 116]. This mechanism supports the use of oral NAC as a defence against neuronal oxidative stress and maintaining redox balance.

Similar to other antioxidants, NAC may exhibit pro-oxidant actions, but these are generally limited to higher doses of 550–1500 mg/kg/day [117–119]. In the current study, NAC was dissolved in drinking water at a concentration of 900 mg/L. Given that an average rat weighing 200 g drinks approximately 40–50 ml/day [120], it would consume 36–45 mg of NAC daily, corresponding to 180–225 mg/kg/day. Importantly, pro-oxidant NAC effects were not observed at concentrations of 50 or 500 mg/kg making pro-oxidative actions of NAC in present study highly unlikely. Nonetheless, optimal doses, treatment schedules and testing efficacy of other antioxidants in preventing THC-induced abnormalities remain to be established. In addition, future studies should explore how NAC (and other antioxidants) may differentially protect the adolescent brain, comparing both female and male experimental cohorts.

The use of a single antioxidant compound in our study is a limitation, but previous research indicated that substituting NAC with other antioxidants can equally prevent behavioral deficits in neurodevelopmental model of schizophrenia [11]. Indeed, several compounds that counteract behavioral aberrations induced by adolescent THC exposure also exhibit antioxidant properties. For example, L-theanine has been shown to reduce oxidative stress by increasing total antioxidant capacity, GSH, and superoxide dismutase [121, 122] and to protect against DA-induced neurotoxicity via boosting GSH levels [123]. Similarly, cannabidiol (CBD), which also directly influences endocannabinoid transmission, diminishes ROS accumulation, activates the Nrf2 pathway through NF-kB inhibition and downregulates oxidative enzymes [124, 125]. CBD also directly influences eCB signalling by reducing CB1 receptor activation, promoting the inverse agonism of CB1 [126], and inhibiting eCB uptake and degradation [127–129]. Of note, CBD has shown behavioural benefits in rats with prenatal THC exposure [130] and in adult female rats with adolescent THC exposure [131]. Ebselen is another promising antioxidant that scavenges ROS and mimics glutathione peroxidase [132]. It has demonstrated neuroprotective effects [133–136] and mimicked NAC’s effects in a neurodevelopmental schizophrenia model [85], supporting the involvement of oxidative stress in the progression of cognitive impairments.